Denis Kokorin1, Jürgen Hennig1, and Maxim Zaitsev1
1Medical Physics, University Medical Center Freiburg, Freiburg, Germany
Synopsis
In this study, the
feasibility of 2D spiral-encoded parallel excitation of limited slice
profiles was investigated. The imaging experiments were performed in
phantoms on a 3T MRI system with 8 RF channels for transmission and
compared to 2D parallel excitation using EPI encoding. The resulting
profiles revealed that 2D spiral-encoded parallel excitation is more
robust against B1 deviations compared to EPI.
Introduction
Application of 2D pulses
has been shown to be advantageous for EPI applications such as DWI
with a reduced FOV in the phase encoding (PE) direction [1-4]. In
this method, multiple thin slice profiles are selected, which are
limited along two encoding directions of the 2D excitation trajectory
and extend across the imaged object in the non-encoded dimension.
Recently, parallel excitation (PEX, [5,6]) has been demonstrated for
shortening 2D EPI-encoded pulses [7-9]. However, application of PEX
to EPI encoding of limited profiles was found suboptimal, since the
undersampling artifacts could not be completely eliminated [7],
especially for high flip angles. In this work, the main goal was to
investigate the feasibility of 2D PEX of limited profiles encoded by
spiral trajectories. For this purpose, we exploited the highest
possible undersampling factors. Finally, we analyzed the benefits and
confounding conditions of 2D PEX of limited slices encoded by spiral
and EPI trajectories.
Theory
In contrast to EPI, the
resolutions along the encoding directions of a spiral trajectory are
related to each other (Figure 1A). The undersampling artifacts for
spirals are represented by swirls and streaks, having little
similarity with the original profile. This effect is due to the fact
that undersampling of the spirals leads to a reduction of the field
of excitation (FOE) along all spatial directions around the excited
voxel and therefore the sample is “aliased” into itself in all
radial dimensions. The undersampling artifacts for EPI are seen as
replicates of the main profile and confined to a narrow location in
space if a thin profile was defined (Figure 1B). In PEX, the
sensitivities of multiple transmit elements are used during 2D pulse
calculation [5] and the undersampling artifacts are eliminated during
excitation. In this manner, artifacts due to small errors and
variations in the B1 data would be blurred over the object for
spirals and their intensity is minimized substantially compared to
EPI trajectories.
Materials and Methods
Experiments were
conducted on a 3T MR scanner (Siemens Magnetom TRIO) with an
8-channel TxArray extension. Slice profiles encoded by spiral
trajectories were excited in a phantom, which was a Plexiglas
container filled with water solution of 1g/L CuSO4 and 0.9 g/L NaCl.
The phantom size was 20×25×30 cm3 and the selected profiles were
imaged using 3D GRE method and 2D SE-EPI. The scanning parameters
were: 1) excitation flip angle = 90°,
FOV = 25.5×25.5 cm2 and matrix = 128×128 for SE-EPI; 2) excitation
flip angle = 30°, FOV =
25.5×19.2×28.8 cm3 and matrix = 128×96×48 for GRE.
The spiral excitation
trajectories were undersampled by factors 2 and 4 (Figure 2). They
were defined on a grid size of 64×64 samples over the FOE of 38.4×24
cm2 and the resulting slice thickness was 12 mm. The 2D pulses were
calculated with an iterative optimization method using conjugate
gradients for the small-tip-angle approximation. Relative axial B1
sensitivities of the transmit coil and an axial B0 map were used in
pulse computations [10].
Results
The selected profiles for
pulses with the undersampling factors of 2 and 4 are demonstrated in
Figures 3 and 4, respectively. Undersampling of excitation trajectories led to
excitation of swirls and streaks re-distributed over the originally
defined FOE as indicated by white arrows in Figures 3A and 4A. Taking into account B1 maps allowed
for suppression of the artifacts to a negligible level comparable to
the image noise. It is important to note that the selected profiles
appeared to be clean from the artifacts along the third non-encoded
dimension of the trajectory. Furthermore, a good precise profile
selection was observed for both excitation flip angles of 30°
and 90°.
Discussion and Summary
In comparison to previous
studies on 2D EPI-encoded parallel excitation of limited profiles
[7,8], spiral pulses are more robust against B1 errors in the
calibration data, which is due to the “spatial incoherence” of
the undersampling artifacts. Another property making spirals
attractive for PEX is that they exhibit slight oversampling at the
k-space center in comparison to peripheral locations, providing
additional encoding information and thus improving further their
robustness against B1 deviations. Nonetheless, spiral pulses are
still relatively long for encoding thin profiles with a thickness
required for clinical applications despite of the demonstrated
shortening with PEX. A remedy could be accomplished by undersampling
spiral trajectories for higher factors, which would require using MRI
systems with a number of transmit channels more than eight [11].
Acknowledgements
The authors would like to thank Dr.
Benjamin Knowles for the helpful discussions.
References
1. P. Mansfield, J. Phys C 10:L5-L58,
1977
2. J. Pauly, J Magn Reson, 82, p. 571,
1989
3. J. Finsterbusch, JMRI, 29, p. 987,
2009
4. EU
Saritas et al., MRM, 60, p. 468, 2008
5. U. Katscher et al.,
MRM, 49, p. 144, 2003
6. Y. Zhu, MRM, 51, p. 775, 2004
7. D. Kokorin et al., Concepts in MR
(B), 45(4), p. 153, 2015
8. D. Kokorin et
al., ISMRM 2015, #2390
9.
D. Kokorin et al., ISMRM 2015, #2941
10.
HP Fautz et al., ISMRM 2008, #1247
11. S.
Schmitter et al., Invest Radiol, 49(5), p. 314, 2014